Paper 2013/112 Original: eprint.iacr.org/2013/112 ↗

PUF Modeling Attacks on Simulated and Silicon Data

Abstract

We show in this paper how several proposed Strong Physical Unclonable Functions (PUFs) can be broken by numerical modeling attacks. Given a set of challenge-response pairs (CRPs) of a Strong PUF, our attacks construct a computer algorithm which behaves indistinguishably from the original PUF on almost all CRPs. This algorithm can subsequently impersonate the PUF, and can be cloned and distributed arbitrarily. This breaks the security of almost all applications and protocols that are based on the respective PUF. The PUFs we attacked successfully include standard Arbiter PUFs and Ring Oscillator PUFs of arbitrary sizes, and XOR Arbiter PUFs, Lightweight Secure PUFs, and Feed-Forward Arbiter PUFs of up to a given size and complexity. The attacks are based upon various machine learning techniques, including a specially tailored variant of Logistic Regression and Evolution Strategies. Our results were obtained on a large number of CRPs coming from numerical simulations, as well as four million CRPs collected from FPGAs and ASICs. The performance on silicon CRPs is very close to simulated CRPs, confirming a conjecture from earlier versions of this work. Our findings lead to new design requirements for secure electrical PUFs, and will be useful to PUF designers and attackers alike.